Classification of maize contaminated withFusarium Graminearum Using Mid-infrared Spectroscopy and Chemometrics

Abstract

Advances in the development of a novel method for the detection ofFusarium graminearum employing mid-infrared spectroscopy are described. The dried and ground sample was sieved and the fraction with particle sizes between 250 and 100 μm was used for spectroscopy. The sample was pressed against a diamond ATR-crystal, mounted in the sample chamber of an FT-IR spectrometer and the mid-infrared absorption spectrum was recorded. The most prominent features in the spectrum were identified as protein, lipid and carbohydrate bands, which are changed by fungal infection. Multivariate data analysis was employed to detect the spectral changes: Principal component analysis (PCA) of mean centred data proved to be the most efficient method for the separation of contaminated maize from uninfected samples using the first two principal components. Ergosterol and the toxin deoxynivalenol (DON) served as reference parameters and were obtained from each sample using conventional analytical techniques. Samples with a toxin content of as low as 309 μg/kg could be separated from blank samples, which is well in the range of natural contamination. Investigated concentration ranges were 0.73-4.5 mg/kg for ergosterol and 0-2.6 mg/kg for DON. The percentage of correctly classified samples was between 75 and 100%.